Abstract

Background: The requirement for hospitalization of patients on dialysis is likely to be a surrogate marker of age and comorbid diseases. It may also reflect the level of care delivered, and substantially increases the cost of this expensive therapy.

Aim: To identify the factors most strongly associated with hospitalization.

Design: Prospective population study.

Methods: Data were recorded for all patients starting RRT in Scotland over one year, including the reasons for and duration of, each hospital admission during the first year of RRT. Factors most strongly associated with hospitalization were determined by Poisson regression analysis.

Results: Overall, 526 patients were admitted to hospital on 1668 occasions (median 3, IQR 1–4) for 13 384 days (median 13, IQR 4–35). Formation of vascular access for haemodialysis (HD) was the most frequent reason for admission, followed by infections. Age, comorbidity, mode of presentation for RRT and primary renal diagnosis were all significantly associated with prolonged hospitalization. Attainment of UK Renal Association standards for urea reduction ratio and serum albumin concentration, and vascular access in the form of arterio-venous fistulae were associated with less hospitalization in patients treated with HD by 90 days.

Discussion: Patients in their first year of RRT have a high requirement for in-patient care, 8.6% of patient treatment days being spent in hospital. Vascular access formation, failure and complications account for a large proportion of this. Age and comorbidity prolong the time spent in hospital. As the RRT population continues to increase, with older patients and those with greater comorbidity, in-patient facilities must also expand.

Introduction

Hospitalization of patients with end-stage renal disease (ESRD) on renal replacement therapy (RRT) substantially increases the cost of this expensive therapy. Hospital admission may be required for the formation of dialysis access, for initiation of RRT, or as a consequence of complications of dialysis or dialysis access, as well as for inter-current illnesses.

The rate and duration of hospitalization are likely to be a surrogate marker of comorbid disease. Increased comorbidity is associated with prolonged duration of hospital admissions in patients without renal failure, an effect which is augmented by age.1 It is therefore not possible to assess how hospitalization reflects the level of care delivered without correcting for comorbidity.

Processes of care have been shown to affect hospitalization in the North American dialysis population: patients treated with recombinant human erythropoietin have fewer hospital admissions due to cardiac, infectious or gastrointestinal symptoms than controls matched for age and cardiovascular history.2 The CANUSA study demonstrated that low serum albumin, poor nutritional state and low creatinine clearance in patients receiving PD were associated with an increase in the hospitalization rate, as well as with mortality.3 Fried and colleagues also found that creatinine clearance was an independent predictor of admissions for patients receiving PD.4

This prospective study of a national incident cohort of patients starting RRT for ESRD aimed to define the burden placed upon hospital in-patient services during the first year of such treatment. In addition, we aimed to determine the proportion of such admissions attributable to direct complications of RRT, and to identify which factors act as predictive indicators of repeated or prolonged hospital admissions.

Methods

Between 1 October 1997 and 30 September 1998, all adult patients starting RRT for ESRD in Scotland's eleven adult renal units were registered when they started RRT. Patients were identified weekly by one of us contacting a liaison person in each unit, to whom a monthly checklist was circulated to ensure no patient had been missed. The data routinely collected by the Scottish Renal Registry (SRR) were augmented with data relating to comorbidity and hospitalization. The data were collected at each site by a single observer (WM) from patients, case notes, and where available, renal unit computer systems, and entered directly onto the Registry via a secure modem. Patients were registered on the day they started RRT, except those failing to recover from presumed acute renal failure, who were registered when the physician caring for them considered they would require permanent RRT. Follow-up data from the same sources were collected at 3-monthly intervals throughout the first year of each patient's treatment. At the time of commencing RRT, the patients’ comorbid illnesses were recorded and each patient was assigned a Charlson comorbidity score.5 For studies with follow-up periods of < 5 years, it is recommended that the scale is used without the addition of an adjustment for age; therefore only comorbidity was taken into account and age was analysed separately. In addition, angina, which is not included in the Charlson index, was graded according to the Canadian Cardiovascular Society index.6

Patients were grouped according to how they presented for RRT: planned, unplanned, unrecovered acute renal failure, acute on chronic renal failure, and end-stage chronic renal failure (Table 1). Patients’ primary renal diagnoses, coded according to the ERA-EDTA codes, were divided into seven groups: primary glomerulonephritis; interstitial nephritis; secondary glomerulonephritis (including vasculitis, SLE, scleroderma); hypertensive nephropathy; multi-system disease (including atheromatous disease, myeloma, amyloid); diabetic nephropathy; unknown cause. These presentation and diagnostic groupings have significant discrimination when analysing mortality within the first 90 days of RRT.7

Table 1

Mode of presentation for RRT

Presentation  
Planned Permanent vascular or peritoneal access ready for use for first RRT 
Unplanned Follow-up by nephrologist for ≥ 1 month, steady progression to end-stage, no permanent access 
Acute renal failure Presentation with presumed acute renal failure but failure to recover function 
Acute on chronic Follow-up by nephrologist for ≥ 1 month, unexpected acute deterioration in function to reach end-stage 
End-stage Require RRT within 1 month of presentation to a nephrologist 
Presentation  
Planned Permanent vascular or peritoneal access ready for use for first RRT 
Unplanned Follow-up by nephrologist for ≥ 1 month, steady progression to end-stage, no permanent access 
Acute renal failure Presentation with presumed acute renal failure but failure to recover function 
Acute on chronic Follow-up by nephrologist for ≥ 1 month, unexpected acute deterioration in function to reach end-stage 
End-stage Require RRT within 1 month of presentation to a nephrologist 

Of 542 patients registered, 16 were excluded from the analysis: one who received a pre-emptive renal transplant and 15 who unexpectedly recovered adequate native renal function to become independent of RRT within a year.

Of the remaining 526, 23 (4.4%) received a successful kidney transplant within the first year of treatment for ESRD. These patients were censored from the analysis at the time of their admission to hospital for their transplant operation. One hundred and forty-four (27.4%) patients died during the first year of treatment and were censored on the date of death.

Very few (< 5%) fistulae were created using PTFE graft material, and therefore PTFE grafts are not differentiated from native vessel fistulae.

Admissions to hospital were counted in whole days, necessarily including an overnight stay, from the day of initiation of RRT for ESRD. Procedures performed as a day case, not necessitating overnight admission, are thus not included in the figures. Days spent in hospital before starting RRT are excluded from the analysis unless otherwise stated. Each admission was coded according to the main reason for the stay in hospital. Hospital re-admissions were calculated as a rate per 100 days at risk; a patient already in hospital could not be re-admitted, therefore the denominator was calculated as the duration of time spent in the community.

Haemoglobin and albumin concentration for all patients, and urea reduction ratio (URR) for patients receiving HD, were recorded at three-monthly intervals from the start of RRT. The attained serum albumin, haemoglobin and URR were calculated as the arithmetical means (for each patient) of those values at each of 3, 6, 9 and 12 months duration of RRT or until death. Examination of the mean values at each time point, both for patients who survived and those who died, showed that there was no long term decline in those patients who died. Therefore the arithmetical mean of results at 3–12 months was a valid representation of the attained standards of treatment over the period of study. Patients were grouped according to whether or not this mean value attained the standard recommended by the United Kingdom Renal Association (UKRA), that is an haemoglobin of at least 10 g/dl, a serum albumin concentration of at least 35 g/l and a URR of 65% or higher.8

Serum albumin was analysed using bromocresol green in all but 17 (3.2%) patients, for whom bromocresol purple was used.

Statistical methods

Data are described using frequencies, percentages, medians and interquartile ranges (IQR). For univariate analyses, χ2 or Kruskal-Wallis tests were used where appropriate to identify factors associated with hospitalization within the first year of RRT. Poisson regression was employed for multivariate analyses. A 5% significance level was used throughout. Statistical analyses were performed using SPSS for Windows (v. 9.0), for univariate analyses and Stata for Windows (v. 6.0) for multivariate Poisson regression analyses.

Results

The median age of patients starting RRT in the year studied (with exclusions given above) was 64.7 years (IQR 51.2–72.0; range 15.6–91.4). The 526 patients studied were admitted to hospital on 1668 occasions (median 3; range 0–18; IQR 1–4) during the first year of RRT. They spent a total of 13 384 (median 13; range 0–358; IQR 4–35) days in hospital from the date of starting RRT for ESRD until death, transplantation or the completion of one year of treatment. Of the 154 779 patient days of RRT during the year of study, 8.6% were spent as a hospital in-patient. Many patients were hospital in-patients prior to commencing RRT for ESRD; in total, 5709 additional days of in-patient care were used prior to the start of RRT for ESRD. Some patients received RRT for presumed acute renal failure during this time, the starting date of RRT for ESRD being some time later when the consultant caring for them judged that recovery was no longer expected.

One hundred and twenty (22.8%) patients started RRT as an out-patient. Those starting RRT as in-patients (n = 406) had a median in-patient stay of 9.5 days following the initiation of RRT for ESRD (range 1–358, IQR 5–19) i.e. excluding time they were judged to have ‘acute’ renal failure. Only 3% of those starting RRT as an out-patient had presented with either acute, acute-on-chronic or end-stage renal failure. In contrast, 43% of those with a planned presentation and 15% with an unplanned presentation commenced RRT with no requirement for hospitalization at the time.

Forty-three patients commenced RRT as in-patients and died during the same hospital admission (8.2% of the cohort and 10.6% of those starting RRT as an in-patient), thus spending 100% of their time on RRT for ESRD in hospital. In total, 2269 in-patient days were accounted for in this way, with patients receiving RRT during 1548 of these days. The median duration of admission (from admission to death) for these patients was 34 days (range 2–441, IQR 22–70); they received RRT for ESRD for a median of 25 days (range 1–358).

The 480 patients who either survived the initiation of RRT to be discharged into the community or started RRT as an out-patient, had a further 1263 hospital admissions during the first year of RRT, median 2, IQR1–4 (range 0–17); median in-patient stay 10 days, IQR 2–26 (range 0–154). Further admissions to hospital after RRT was started were most common during the first three months of RRT but continued thereafter at a steady rate for the whole first year of treatment (Figure 1).

Figure 1.

Frequency of admissions to hospital by month of RRT.

Figure 1.

Frequency of admissions to hospital by month of RRT.

The most common single reason for admission was the creation of vascular access, followed by both general and access infections (Table 2). The admissions grouped as ‘other cause’ were heterogeneous, including time in hospital due to post dialysis complications such as hypotension, gastrointestinal symptoms (e.g. diarrhoea and vomiting) and gastro-intestinal investigations. In addition, time spent in hospital for terminal care and social reasons are included here; no single reason alone accounts for a large proportion of the time.

Table 2

Admissions to hospital in the first year of RRT

Reason for admission No. admissions (% of total) Total duration (days) (% of total) Admission duration
 
  
   Median IQR Range 
Vascular access creation 181 (14.3%) 511 (5.3%) 2–3 1–18 
Line/permcath infection 108 (8.6%) 901 (9.4%) 3–8.8 1–121 
Access complication 72 (5.7%) 320 (3.3%) 1–4 1–85 
Tenckhoff insertion 26 (2.1%) 90 (0.9%) 2–4 1–12 
PD training 30 (2.4%) 313 (3.3%) 7.5 3–17.3 2–39 
PD peritonitis 80 (6.3%) 587 (6.1%) 3–9 1–41 
Fluid overload 81 (6.4%) 434 (4.5%) 2–7 1–40 
Surgery 77 (6.1%) 944 (9.8%) 2–11.5 1–137 
Vascular event 71 (5.6%) 815 (8.5%) 2–12 1–88 
Other infection 114 (9.0%) 1275 (13.3%) 3–14 1–97 
Other 423 (33.5%) 3423 (35.6%) 2–10 1–146 
Total 1263 9613    
Reason for admission No. admissions (% of total) Total duration (days) (% of total) Admission duration
 
  
   Median IQR Range 
Vascular access creation 181 (14.3%) 511 (5.3%) 2–3 1–18 
Line/permcath infection 108 (8.6%) 901 (9.4%) 3–8.8 1–121 
Access complication 72 (5.7%) 320 (3.3%) 1–4 1–85 
Tenckhoff insertion 26 (2.1%) 90 (0.9%) 2–4 1–12 
PD training 30 (2.4%) 313 (3.3%) 7.5 3–17.3 2–39 
PD peritonitis 80 (6.3%) 587 (6.1%) 3–9 1–41 
Fluid overload 81 (6.4%) 434 (4.5%) 2–7 1–40 
Surgery 77 (6.1%) 944 (9.8%) 2–11.5 1–137 
Vascular event 71 (5.6%) 815 (8.5%) 2–12 1–88 
Other infection 114 (9.0%) 1275 (13.3%) 3–14 1–97 
Other 423 (33.5%) 3423 (35.6%) 2–10 1–146 
Total 1263 9613    

During the first year of RRT for ESRD a total of 144 (27.4%) patients died, median survival 81 days, IQR 36–220. Seventy-three of these patients died within the first 90 days of treatment. One-year mortality, excluding 90 days deaths, was 15.8%.9 For each patient, we calculated the proportion of the time spent as a hospital in patient from the start of RRT until the completion of one year of RRT, death or renal transplantation.

Univariate analysis revealed that increasing age at the time of starting RRT, primary renal diagnosis, burden of comorbid illnesses, diabetes, non-planned presentation for RRT and the use of temporary vascular access for HD were associated with both prolonged hospitalization and with repeated admissions. Patients initially treated with HD rather than PD spent longer in hospital and were more likely to be admitted. This difference was maintained after 3 months of RRT. By the end of 90 days of treatment, the use of either temporary central venous cannulae or permcaths as vascular access for HD were again associated with worse outcomes. Attaining recommended standards for haemoglobin and albumin concentrations, and urea reduction ratio on HD were associated with favourable outcomes (Table 3). Patients’ sex, socioeconomic status, smoking habit and calculated (Cockcroft and Gault10) residual renal function at the time of starting RRT, had no effect upon the proportion of time spent in hospital nor upon re-admissions during the first year of RRT.

Table 3

Univariate analysis: factors affecting proportion of time on RRT spent in hospital and number of re-admissions to hospital (per 100 days at risk)

Variable n (526) %RRT in hospital
 
 p n (483) Re-admissions
 
 p 
  Median IQR   Median IQR  
Age group    < 0.001    < 0.001 
15–49 years 119 2.0 0.8–5.4  115 0.6 0.3–1.1  
50–64 years 150 3.4 0.8–11.9  147 0.6 0.3–1.4  
65–75 years 170 9.3 2.4–25.1  150 0.9 0.3–2.2  
> 75 years 87 14.5 5.2–35.1  71 1.3 0.6–2.5  
Renal diagnosis    < 0.001    < 0.001 
Glomerulonephritis 69 2.7 0.3–10.0  66 0.6 0–1.3  
Interstitial nephritis 121 2.2 0.6–7.1  119 0.6 0.3–0.9  
Diabetic 85 8.4 2.6–22.8  77 1.1 0.6–2.6  
Multi-system 79 16.2 6.0–68.4  64 1.3 0.6–2.7  
Hypertension 21 3.6 1.0–7.5  20 0.3 0.3–0.9  
Secondary GN 27 9.3 0.8–21.3  26 1.0 0–3.0  
Unknown 124 5.3 2.2–19.4  111 0.6 0.3–1.8  
Charlson score    < 0.001    < 0.001 
195 2.0 0.3–4.9  188 0.3 0–0.8  
104 7.4 1.9–19.0  98 0.9 0.3–2.2  
122 11.0 2.7–26.1  107 1.2 0.6–2.5  
50 8.1 2.7–24.6  44 1.1 0.4–2.1  
37 16.4 3.8–42.5  30 1.6 0.6–2.5  
13 21.1 7.5–79.8  11 1.7 –3.7  
10.9 1.8–26.2  1.8 0.3–4.0  
Diabetes    0.04    < 0.001 
Yes 123 8.4 2.7–22.2  111 1.2 0.6–2.3  
No 403 4.4 1.1–16.2  372 0.6 0.3–1.5  
Angina    0.01    0.002 
Grade 0 374 4.1 1.1–15.1  350 0.6 0.3–1.6  
Grade I 67 8.8 1.7–43.5  56 0.7 0.3–2.1  
Grade II 62 8.1 2.7–20.5  55 1.1 0.3–2.6  
Grade III 18 11.2 3.7–19.7  17 1.6 0.6–2.3  
Grade IV 23.0 4.5–55.8  2.7 1.4–5.3  
Presentation    < 0.001    < 0.001 
Planned 226 1.9 0–5.4  223 0.4 0–0.9  
Unplanned 129 6.3 2.6–15.3  122 1.0 0.6–2.1  
ARF 53 21.6 11.4–8.3  43 2.3 1.0–3.1  
Acute on CRF 62 19.6 6.2–89.9  47 1.3 0.6–2.5  
ESRD 56 7.7 2.7–42.7  48 0.9 0.3–2.4  
Initial mode RRT    0.05    < 0.001 
HD 402 6.6 1.9–21.4  366 0.9 0.3–2.1  
PD 124 2.7 1.1–7.4  117 0.6 0–0.9  
Initial HD access n = 401   < 0.001 n = 365   < 0.001 
Line 276 11.6 4.1–28.4  243 1.2 0.6–2.6  
Permcath 30 12.1 2.7–22.7  29 1.4 0.7–2.0  
Fistula/graft 95 0.5 0–2.7  93 0.3 0–0.6  
90-day modality    0.06    0.02 
HD 295 4.4 0.8–13.2  293 0.8 0.3–1.6  
PD 151 3.0 1.1–7.6  151 0.6 0.3–1.2  
90-day HD access    < 0.001    <0.001 
Line 69 8.8 2.9–19.5  69 1.6 0.8–2.5  
Permcath 72 11.7 5.5–19.4  70 1.1 0.6–1.8  
Fistula/graft 154 1.6 0–5.0  154 0.4 0–0.9  
Average URR    < 0.001    < 0.001 
< 65 137 7.4 1.9–16.6  136 1.1 0.5–2.1  
≥ 65 171 3.0 0.5–8.0  171 0.6 0.3–1.2  
Average Hb    < 0.001    < 0.001 
< 10 165 7.1 2.7–16.7  163 1.1 0.6–2.0  
> 10 282 2.6 0.5–7.1  282 0.6 0.3–1.2  
Average albumin    < 0.001    < 0.001 
< 35 g/l 111 14.0 4.1–23.0  109 1.5 0.6–2.6  
> 35 g/l 335 2.7 0.8–6.3  335 0.6 0.3–1.1  
Variable n (526) %RRT in hospital
 
 p n (483) Re-admissions
 
 p 
  Median IQR   Median IQR  
Age group    < 0.001    < 0.001 
15–49 years 119 2.0 0.8–5.4  115 0.6 0.3–1.1  
50–64 years 150 3.4 0.8–11.9  147 0.6 0.3–1.4  
65–75 years 170 9.3 2.4–25.1  150 0.9 0.3–2.2  
> 75 years 87 14.5 5.2–35.1  71 1.3 0.6–2.5  
Renal diagnosis    < 0.001    < 0.001 
Glomerulonephritis 69 2.7 0.3–10.0  66 0.6 0–1.3  
Interstitial nephritis 121 2.2 0.6–7.1  119 0.6 0.3–0.9  
Diabetic 85 8.4 2.6–22.8  77 1.1 0.6–2.6  
Multi-system 79 16.2 6.0–68.4  64 1.3 0.6–2.7  
Hypertension 21 3.6 1.0–7.5  20 0.3 0.3–0.9  
Secondary GN 27 9.3 0.8–21.3  26 1.0 0–3.0  
Unknown 124 5.3 2.2–19.4  111 0.6 0.3–1.8  
Charlson score    < 0.001    < 0.001 
195 2.0 0.3–4.9  188 0.3 0–0.8  
104 7.4 1.9–19.0  98 0.9 0.3–2.2  
122 11.0 2.7–26.1  107 1.2 0.6–2.5  
50 8.1 2.7–24.6  44 1.1 0.4–2.1  
37 16.4 3.8–42.5  30 1.6 0.6–2.5  
13 21.1 7.5–79.8  11 1.7 –3.7  
10.9 1.8–26.2  1.8 0.3–4.0  
Diabetes    0.04    < 0.001 
Yes 123 8.4 2.7–22.2  111 1.2 0.6–2.3  
No 403 4.4 1.1–16.2  372 0.6 0.3–1.5  
Angina    0.01    0.002 
Grade 0 374 4.1 1.1–15.1  350 0.6 0.3–1.6  
Grade I 67 8.8 1.7–43.5  56 0.7 0.3–2.1  
Grade II 62 8.1 2.7–20.5  55 1.1 0.3–2.6  
Grade III 18 11.2 3.7–19.7  17 1.6 0.6–2.3  
Grade IV 23.0 4.5–55.8  2.7 1.4–5.3  
Presentation    < 0.001    < 0.001 
Planned 226 1.9 0–5.4  223 0.4 0–0.9  
Unplanned 129 6.3 2.6–15.3  122 1.0 0.6–2.1  
ARF 53 21.6 11.4–8.3  43 2.3 1.0–3.1  
Acute on CRF 62 19.6 6.2–89.9  47 1.3 0.6–2.5  
ESRD 56 7.7 2.7–42.7  48 0.9 0.3–2.4  
Initial mode RRT    0.05    < 0.001 
HD 402 6.6 1.9–21.4  366 0.9 0.3–2.1  
PD 124 2.7 1.1–7.4  117 0.6 0–0.9  
Initial HD access n = 401   < 0.001 n = 365   < 0.001 
Line 276 11.6 4.1–28.4  243 1.2 0.6–2.6  
Permcath 30 12.1 2.7–22.7  29 1.4 0.7–2.0  
Fistula/graft 95 0.5 0–2.7  93 0.3 0–0.6  
90-day modality    0.06    0.02 
HD 295 4.4 0.8–13.2  293 0.8 0.3–1.6  
PD 151 3.0 1.1–7.6  151 0.6 0.3–1.2  
90-day HD access    < 0.001    <0.001 
Line 69 8.8 2.9–19.5  69 1.6 0.8–2.5  
Permcath 72 11.7 5.5–19.4  70 1.1 0.6–1.8  
Fistula/graft 154 1.6 0–5.0  154 0.4 0–0.9  
Average URR    < 0.001    < 0.001 
< 65 137 7.4 1.9–16.6  136 1.1 0.5–2.1  
≥ 65 171 3.0 0.5–8.0  171 0.6 0.3–1.2  
Average Hb    < 0.001    < 0.001 
< 10 165 7.1 2.7–16.7  163 1.1 0.6–2.0  
> 10 282 2.6 0.5–7.1  282 0.6 0.3–1.2  
Average albumin    < 0.001    < 0.001 
< 35 g/l 111 14.0 4.1–23.0  109 1.5 0.6–2.6  
> 35 g/l 335 2.7 0.8–6.3  335 0.6 0.3–1.1  

URR, urea reduction ratio.

Under multivariate Poisson regression, prolonged duration of hospital admission was associated with age, comorbidity, mode of presentation for RRT and primary renal diagnosis group (Table 4). Patients aged > 75 years spent 95% longer in hospital than those aged < 50 years, and each one point increment on the Charlson index was associated with a 14% longer stay in hospital. Those patients with ESRD due to diabetic nephropathy spent 73% longer in hospital than those with primary glomerulonephritis (GN) and those with multisystem disease 52% longer. A primary renal diagnosis of hypertensive nephropathy or secondary GN made no significant difference to hospitalization duration compared with primary GN. Patients who had been followed up by a nephrologist prior to receiving RRT spent 1% less time in hospital for each month of pre-dialysis care. When analysed taking into account all other factors, the use of HD, as first mode of RRT, was not associated with prolonged admissions. possibly because the differences seen in the univariate analysis were actually due to differences in age and comorbidity. It was however associated with an increased number of admissions, even after adjusting for these other factors. Therefore the use of HD was associated with decreased number of days spent in hospital but an increased number of admissions compared with PD (Tables 4 and 6).

Table 4

Poisson regression multivariate analysis: duration of hospital admissions from the start of RRT

Variable Incidence rate ratio 95%CI p 
Age group    
15–49 years 1.0 – – 
50–64 years 1.26 1.19–1.33 < 0.001 
65–75 years 1.52 1.44–1.61 < 0.001 
> 75 years 1.95 1.84–2.07 < 0.001 
    
Charlson score 1.14 1.13–1.16 < 0.001 
Angina 1.06 1.02–1.10 0.004 
Follow-up 0.99 0.997–0.998 < 0.001 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 1.15 1.07–1.23 < 0.001 
Multi-system 1.52 1.41–1.63 < 0.001 
Secondary GN 1.0 0.91–1.10 0.9 
Hypertension 1.0 0.91–1.12 0.9 
Diabetic 1.73 1.60–1.87 < 0.001 
Unknown 1.12 1.05–1.21 0.001 
    
Presentation    
Planned 1.0 – – 
Unplanned 1.78 1.68–1.89 < 0.001 
Acute renal failure 4.87 4.57–5.19 < 0.001 
Acute on CRF 2.57 2.42–2.74 < 0.001 
ESRD 2.40 2.25–2.56 < 0.001 
    
Initial mode    
PD 1.0 – – 
HD 0.91 0.86–0.96 0.001 
Variable Incidence rate ratio 95%CI p 
Age group    
15–49 years 1.0 – – 
50–64 years 1.26 1.19–1.33 < 0.001 
65–75 years 1.52 1.44–1.61 < 0.001 
> 75 years 1.95 1.84–2.07 < 0.001 
    
Charlson score 1.14 1.13–1.16 < 0.001 
Angina 1.06 1.02–1.10 0.004 
Follow-up 0.99 0.997–0.998 < 0.001 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 1.15 1.07–1.23 < 0.001 
Multi-system 1.52 1.41–1.63 < 0.001 
Secondary GN 1.0 0.91–1.10 0.9 
Hypertension 1.0 0.91–1.12 0.9 
Diabetic 1.73 1.60–1.87 < 0.001 
Unknown 1.12 1.05–1.21 0.001 
    
Presentation    
Planned 1.0 – – 
Unplanned 1.78 1.68–1.89 < 0.001 
Acute renal failure 4.87 4.57–5.19 < 0.001 
Acute on CRF 2.57 2.42–2.74 < 0.001 
ESRD 2.40 2.25–2.56 < 0.001 
    
Initial mode    
PD 1.0 – – 
HD 0.91 0.86–0.96 0.001 

Those patients who were receiving HD by 90 days of treatment (n = 295) were analysed separately (Table 5). Again, increasing age and burden of comorbidity adversely affected the proportion of time spent in hospital, those aged > 75 years spending 60% longer in hospital than the under-50s. Each increment in the Charlson score added 9% to duration of hospitalization. Those presenting with unrecovered acute renal failure spent significantly longer (14%) in hospital than the planned group. The impact of attaining vascular access for HD in the form of a fistula rather than temporary line or permcath is marked. Those patients with a temporary line at the time of initiation of RRT spending 3.7 times as long in hospital, and those still dialysing with a line by 90 days spending 1.7 times as long as those with fistulae. Patients attaining the UKRA recommended standards for albumin; haemoglobin and URR spent 53%, 22% and 10% less time in hospital, respectively, than those failing to attain the standards.

Table 5

Poisson regression multivariate analysis: duration of hospital admissions from the start of RRT of those on HD at 90 days

Variable Incidence rate ratio 95%CI p 
Age group    
15–49 years 1.0 – – 
50–64 years 1.41 1.30–1.54 < 0.001 
65–75 years 1.39 1.28–1.51 < 0.001 
> 75 years 1.60 1.46–1.75 < 0.001 
    
Charlson score 1.09 1.06–1.11 < 0.001 
Follow-up 0.998 0.997–0.999 0.004 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 1.41 1.28–1.54 < 0.001 
Multi-system 1.67 1.51–1.84 < 0.001 
Secondary GN 1.16 1.00–1.34 0.05 
Hypertension 0.83 0.72–0.94 0.004 
Diabetic 1.53 1.37–1.70 < 0.001 
Unknown 1.60 1.44–1.77 < 0.001 
    
Presentation    
Planned 1.0 – – 
Unplanned 0.65 0.57–0.75 < 0.001 
Acute renal failure 1.14 0.98–1.31 0.09 
Acute on CRF 0.70 0.60–0.81 < 0.001 
ESRD 0.62 0.53–0.72 < 0.001 
    
Initial access    
Fistula 1.0 – – 
Line 3.67 3.12–4.32 < 0.001 
Permcath 2.87 2.53–3.25 < 0.001 
Tenckhoff for PD 5.92 5.12–6.86 < 0.001 
    
Access at 3 months    
Fistula 1.0 – – 
Line 1.71 1.61–1.84 < 0.001 
Permcath 1.96 1.83–2.11 < 0.001 
    
Average albumin > 35 g/l 0.47 0.45–0.50 < 0.001 
Average haemoglobin > 10 g/dl 0.78 0.74–0.82 < 0.001 
Average URR > 65% 0.90 0.85–0.95 < 0.001 
Variable Incidence rate ratio 95%CI p 
Age group    
15–49 years 1.0 – – 
50–64 years 1.41 1.30–1.54 < 0.001 
65–75 years 1.39 1.28–1.51 < 0.001 
> 75 years 1.60 1.46–1.75 < 0.001 
    
Charlson score 1.09 1.06–1.11 < 0.001 
Follow-up 0.998 0.997–0.999 0.004 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 1.41 1.28–1.54 < 0.001 
Multi-system 1.67 1.51–1.84 < 0.001 
Secondary GN 1.16 1.00–1.34 0.05 
Hypertension 0.83 0.72–0.94 0.004 
Diabetic 1.53 1.37–1.70 < 0.001 
Unknown 1.60 1.44–1.77 < 0.001 
    
Presentation    
Planned 1.0 – – 
Unplanned 0.65 0.57–0.75 < 0.001 
Acute renal failure 1.14 0.98–1.31 0.09 
Acute on CRF 0.70 0.60–0.81 < 0.001 
ESRD 0.62 0.53–0.72 < 0.001 
    
Initial access    
Fistula 1.0 – – 
Line 3.67 3.12–4.32 < 0.001 
Permcath 2.87 2.53–3.25 < 0.001 
Tenckhoff for PD 5.92 5.12–6.86 < 0.001 
    
Access at 3 months    
Fistula 1.0 – – 
Line 1.71 1.61–1.84 < 0.001 
Permcath 1.96 1.83–2.11 < 0.001 
    
Average albumin > 35 g/l 0.47 0.45–0.50 < 0.001 
Average haemoglobin > 10 g/dl 0.78 0.74–0.82 < 0.001 
Average URR > 65% 0.90 0.85–0.95 < 0.001 

URR, urea reduction ratio.

When the rates of admission to hospital, rather than duration of hospital stay, were analysed by Poisson multivariate analysis, comorbidity, initial use of HD and non-planned presentation were associated with increased admission rate (Table 6). Diagnoses of hypertensive nephropathy and interstitial nephritis were associated with significantly fewer admissions than primary GN; patients with diabetic nephropathy, multisystem disease and secondary GN had significantly more admissions. Age was not shown to have a significant effect when analysed in the presence of all other factors. While increased follow-up significantly decreased re-admissions, clinically the impact was minimal, with each extra month of follow-up reducing re-admissions by 0.003%.

Table 6

Poisson regression multivariate analysis: re-admissions of patients from the start of RRT

Variable Incidence rate ratio 95%CI 
Charlson score 1.13 1.08–1.19 < 0.001 
Angina 1.17 1.03–1.33 0.02 
Follow-up 0.997 0.995–0.999 0.001 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 0.91 0.74–1.12 0.4 
Multi-system 1.25 0.99–1.57 0.06 
Secondary GN 1.30 0.99–1.71 0.06 
Hypertension 0.69 0.48–0.99 0.04 
Diabetic 1.31 1.04–1.67 0.02 
Unknown 1.09 0.89–1.35 0.4 
    
Presentation    
Planned 1.0 – – 
Unplanned 1.50 1.30–1.76 < 0.001 
Acute renal failure 2.35 1.90–2.91 < 0.001 
Acute on CRF 1.50 1.22–1.84 < 0.001 
ESRD 1.40 1.13–1.73 0.002 
    
Initial mode    
PD 1.0 – – 
HD 1.27 1.07–1.51 0.006 
Variable Incidence rate ratio 95%CI 
Charlson score 1.13 1.08–1.19 < 0.001 
Angina 1.17 1.03–1.33 0.02 
Follow-up 0.997 0.995–0.999 0.001 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 0.91 0.74–1.12 0.4 
Multi-system 1.25 0.99–1.57 0.06 
Secondary GN 1.30 0.99–1.71 0.06 
Hypertension 0.69 0.48–0.99 0.04 
Diabetic 1.31 1.04–1.67 0.02 
Unknown 1.09 0.89–1.35 0.4 
    
Presentation    
Planned 1.0 – – 
Unplanned 1.50 1.30–1.76 < 0.001 
Acute renal failure 2.35 1.90–2.91 < 0.001 
Acute on CRF 1.50 1.22–1.84 < 0.001 
ESRD 1.40 1.13–1.73 0.002 
    
Initial mode    
PD 1.0 – – 
HD 1.27 1.07–1.51 0.006 

Use of a fistula as access for HD both at the initiation of RRT and by 90 days was associated with a significant reduction in admissions compared with temporary lines or permcaths. (Table 7). When those treated by HD at 90 days were analysed separately, attainment of UKRA standards for albumin and URR reduced admissions by 32% and 20% respectively, but attaining the standard for haemoglobin did not have a significant effect upon readmission rate. Again, age was not a significant factor in re-admission rates when using multivariate analysis to adjust for other factors.

Table 7

Poisson regression multivariate analysis: re-admissions of those on HD at 90 days

Variable Incidence rate ratio 95%CI 
Charlson score 1.10 1.04–1.17 0.001 
Angina 1.29 1.10–1.51 0.002 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 0.87 0.67–1.13 0.3 
Multi-system 1.33 1.01–1.75 0.05 
Secondary GN 1.14 0.78–1.67 0.5 
Hypertension 0.66 0.43–0.99 0.05 
Diabetic 1.17 0.88–1.56 0.3 
Unknown 1.06 0.82–1.37 0.7 
    
Initial access    
Fistula 1.0 – – 
Line 1.64 1.34–2.02 < 0.001 
Permcath 1.74 1.26–2.40 0.001 
Tenckhoff for PD 1.63 1.0–2.65 0.05 
    
Access at 3 months    
Fistula 1.0 – – 
Line 2.03 1.69–2.42 < 0.001 
Permcath 1.42 1.14–1.76 0.002 
    
Average albumin > 35 g/l 0.68 0.59–0.80 < 0.001 
Average URR > 65% 0.80 0.69–0.92 0.002 
Variable Incidence rate ratio 95%CI 
Charlson score 1.10 1.04–1.17 0.001 
Angina 1.29 1.10–1.51 0.002 
    
Renal diagnosis    
Glomerulonephritis 1.0 – – 
Interstitial nephritis 0.87 0.67–1.13 0.3 
Multi-system 1.33 1.01–1.75 0.05 
Secondary GN 1.14 0.78–1.67 0.5 
Hypertension 0.66 0.43–0.99 0.05 
Diabetic 1.17 0.88–1.56 0.3 
Unknown 1.06 0.82–1.37 0.7 
    
Initial access    
Fistula 1.0 – – 
Line 1.64 1.34–2.02 < 0.001 
Permcath 1.74 1.26–2.40 0.001 
Tenckhoff for PD 1.63 1.0–2.65 0.05 
    
Access at 3 months    
Fistula 1.0 – – 
Line 2.03 1.69–2.42 < 0.001 
Permcath 1.42 1.14–1.76 0.002 
    
Average albumin > 35 g/l 0.68 0.59–0.80 < 0.001 
Average URR > 65% 0.80 0.69–0.92 0.002 

URR, urea reduction ratio.

Discussion

This prospective national study demonstrates the high requirement for in-patient care of patients during their first year of RRT, 8.6% of patient days were spent in hospital. Hospitalization was required by 77% of patients at the time of starting RRT when a median of 9.5 days were spent in hospital.

Once established on RRT in the community, creation of vascular access for HD along with infectious and other complications of such access accounted for 28.6% of further admissions to hospital and 18% of days spent in hospital. Almost 10% of days spent in hospital were for other surgical procedures; nine patients who required 11 lower limb amputations accounted for 78% of this time. One third of admissions and just over one third of days spent in hospital were accounted for by miscellaneous reasons such as hypotension, diarrhoea, gastrointestinal bleeding, various investigations, social reasons and terminal care, reflecting the relative fragility of this group of patients.

There was a major impact of comorbidity upon RRT patients’ requirement for in-patient care. Under multivariate analysis, for each additional single point on the Charlson comorbidity index, patients spent 14% longer in hospital and were admitted 13% more frequently. The Charlson index has previously been shown to be independently associated with length of hospital stay for patients admitted with acute chest pain11 and after stroke.12

Age also has a major impact upon duration of hospitalization with those aged > 75 years spending almost twice as long in hospital as the under-50s, despite no significant difference in the number of admissions required. These factors have major implications for future provision of in-patient facilities, since the continuing expansion in the RRT population is largely due to greater numbers of older patients, and those with comorbid illnesses receiving RRT. SRR data show that in 1985–89, only 3% of the incident population were aged >75 years, rising to 15% in 1995–99.13

Patients with ESRD due to diabetic nephropathy spent 73% longer in hospital than those with primary glomerulonephritis; those with multisystem disease spent 52% longer in hospital. SRR data show that the proportion of patients starting RRT due to diabetic nephropathy continues to expand from 3% in 1975–79 to 18% in 1995–99, with no evidence of slowing in this trend.13

The effect of mode of presentation for RRT upon the requirement for hospitalization during the first year of RRT was dramatic. Those patients presenting with unrecovered ‘acute’ renal failure spent almost five times as long in hospital and were re-admitted more than twice as often, than those who had been cared for by a nephrologist for at least one month prior to commencing RRT, and had access ready for use on the day of starting RRT (planned group). Those presenting with an acute deterioration of established chronic renal failure spent almost 2.6 times as long in hospital and were re-admitted 1.5 times as often, and those presenting out of the blue with end-stage renal failure spent 2.4 times as long in hospital and had 1.4 times the number of admissions. The unplanned group who had been followed by a nephrologist for at least a month prior to starting RRT but did not have mature access ready for use at the time of starting spent almost 1.8 times as long in hospital as the planned group and were re-admitted 1.5 times as often.

While some of this time may have been spent in attaining access, this represents a difference over 1 year of approximately 8 extra days in hospital, far longer than one would expect an uncomplicated admission to create access to take. It was beyond the scope of this study to record time spent in hospital by the planned group prior to starting RRT. Previous studies show dialysis access problems to account for between 25–48% of admissions of patients receiving HD, albeit for a far smaller proportion of the actual days spent in hospital.14–16 Aoroa and colleagues also found that delayed referral to nephrology services, and the subsequent lack of permanent vascular access, were independently associated with increased hospitalization within the first three months of RRT.17 The beneficial effect of having optimal access for HD, in the form of an arterio-venous fistula, ready for use by the time the patient starts RRT is emphasized by the fact that those in this study starting RRT with a temporary central venous cannula or semi-permanent tunnelled cannula spent 3.7 and 2.9 times, respectively, as long in hospital as the group with mature fistulae at the time of starting RRT. They also were admitted with greater frequency than the group of patients with fistulae. This effect is still present by 90 days of treatment in the group receiving HD, and is reflected by the large proportion of time spent in hospital and readmissions to hospital due to access formation and complications. The median stay in hospital for vascular access creation was 2 days; this reflects the practice of admitting patients to hospital the day before their procedure and discharging the day after, thus necessitating two nights in hospital. Practice in the UK is moving more towards day case access procedures for those who are well enough. The benefit of attaining access prior to starting RRT may be magnified, because it is in the patients with less comorbidity that access is most readily established.

Previous studies have found the use of grafts rather than fistulae to be associated with increased hospitalization.18 We had very few patients with grafts rather than native vessel fistulae, and they were analysed together. The low use of synthetic grafts in our incident patients is entirely in keeping with the findings of the DOPPS study, which shows the UK, in common with the rest of Europe, to have a much lower use of grafts than in the USA.19

Patients who were receiving HD by 90 days of treatment and attained the recommended UKRA standards for urea reduction ratio, haemoglobin and albumin concentrations were admitted for less time and on fewer occasions than those who failed to attain the recommended standards. This effect was found by multivariate analysis, with comorbidity and access taken into consideration, thus the possible impact of poor access and high comorbidity on attainment of recommended standards has been adjusted for. The use of catheters as HD access is known to be a barrier to optimal delivery of HD,20 but both access for HD and URR had a significant effect upon hospitalization of those treated by HD after 90 days. Dialysis adequacy is known to be an important factor in survival of patients receiving RRT,2,21,22 and therefore our finding of a beneficial association between dialysis adequacy and a reduction in morbidity and hospitalization rates is as expected.

Low serum albumin has previously been found to be associated with increased hospitalization in patients receiving both HD18 and PD2,3 as well as with mortality.21,23 Anaemia has also been previously demonstrated to be associated with increased hospitalization in the pre-dialysis population24 as well as those treated by RRT.25,26 To our knowledge however this is the first study which has demonstrated a positive association with hospitalization rates, of achieving the UKRA recommended standards for albumin, URR and haemoglobin.

The population starting RRT continues to expand, with older patients and those with comorbid illnesses. During the first year of RRT, elderly patients, those with high comorbidity burdens and those who present for RRT in an unprepared manner starting treatment with temporary access, place large pressures upon hospital in-patient services. The challenge is to improve the morbidity, as measured by hospitalization, of patients treated by RRT by improving the standards of delivered care. Emphasis must be placed upon attaining optimal access for dialysis as quickly as possible and achieving the recommended UKRA standards for anaemia, albumin and dialysis adequacy. However, it seems inevitable that in-patient facilities must expand as the RRT population becomes more aged and infirm.

We would like to thank the staff of Scotland's adult renal units for their help in executing this study: Aberdeen Royal Infirmary; Crosshouse Hospital, Kilmarnock; Dumfries and Galloway Royal Infirmary; Glasgow Royal Infirmary; Monklands Hospital, Airdrie; Ninewells Hospital, Dundee; Queen Margaret's Hospital, Dunfermline; Raigmore Hospital, Inverness; Royal Infirmary of Edinburgh; Stobhill Hospital, Glasgow; Western Infirmary, Glasgow. This study was funded by a grant from the Scottish Office Department of Health Clinical Resource and Audit Group (CRAG). The views expressed are those of the authors and not necessarily those of either the Scottish Office Department of Health or CRAG. The Scottish Renal Registry at the time of the study was funded jointly by all the Scottish Health Boards.

References

1
Rochon PA, Katz JN, Morrow LA, McGlinchey-Berroth R, Ahlquist MM, Sarkarati M, Minaker KL. Comorbid illness is associated with survival and length of hospital stay in patients with chronic disability.
Med Care
 
1996
;
34
:
1093
–101.
2
Churchill DN, Muirhead N, Goldstein M, Posen G, Fay W, Beecroft ML, Gorman J, Taylor DW. Effect of recombinant human erythropoietin on hospitalisation of hemodialysis patients.
Clin Nephrol
 
1995
;
43
:
184
–8.
3
Canada-USA (CANUSA) peritoneal dialysis study group. Adequacy of dialysis and nutrition in continuous peritoneal dialysis: Association with clinical outcomes.
J Am Soc Nephrol
 
1996
;
7
:
198
–207.
4
Fried L, Abidi S, Bernardini J, Johnston JR, Piraino B. Hospitalisation in peritoneal dialysis patients.
Am J Kidney Dis
 
1999
;
33
:
927
–33.
5
Charlson ME, Pompei P, Ales KL, MacKenzie RC. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation.
J Chron Dis
 
1987
;
40
:
373
–83.
6
Campeau L. Grading of Angina Pectoris.
Circulation
 
1976
;
54
:
522
–3.
7
Metcalfe W, Khan IH, Prescott GJ, Simpson K, MacLeod AM, on behalf of the Scottish Renal Registry. Can we improve early mortality in patients receiving renal replacement therapy?
Kidney Int
 
2000
;
57
:
2539
–45.
8
Renal Association standards subcommittee.
Treatment of adult patients with renal failure: Recommended standards and audit measures
 , 2nd edn. London, Royal College of Physicians,
1997
.
9
Metcalfe W, Khan IH, Prescott GJ, Simpson K, MacLeod AM, on behalf of the Scottish Renal Registry. End-stage renal disease in Scotland: Outcomes and standards of care.
Kidney Int
 
2003
;
64
:
1808
–16.
10
Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine.
Nephron
 
1976
;
16
:
31
–41.
11
Matsui K, Goldman L, Johnson PA, Kuntz KM, Cook EF, Lee TH. Comorbidity as a correlate of length of stay for hospitalized patients with acute chest pain.
J Gen Intern Med
 
1996
;
11
:
262
–8.
12
Monane M, Kanter DS, Glynn RJ, Avorn J. Variability in length of hospitalisation for stroke: The role of managed care in an elderly population.
Arch Neurol
 
1996
;
53
:
875
–80.
13
Scottish Renal Association.
The Scottish Renal Registry Second Annual Report, 1999.
  Edinburgh, Information and Statistics Division,
2001
.
14
Carlson DM, Duncan DA, Naessens JM, Johnson WJ. Hospitalisation in dialysis patients.
Mayo Clin Proc
 
1984
;
59
:
769
–75.
15
Morduchowicz G, Boner G. Hospitalisation in dialysis end-stage renal failure patients.
Nephron
 
1996
;
73
:
413
–16.
16
Ifudu O, Mayers JD, Cohen LS, Paul H, Brezsnyak WF, Avram MM, Herman AI, Friedman EA. Correlates of vascular access and non-vascular access-related hospitalisations in haemodialysis patients.
Am J Nephrol
 
1996
;
16
:
118
–23.
17
Arora P, Kausz AT, Obrador GT, Ruthazer R, Khan S, Jenuleson CS, Meyer KB, Pereira BJG. Hospital utilization among chronic dialysis patients.
J Am Soc Nephrol
 
2000
;
11
:
740
–6.
18
Churchill DN, Taylor DW, Cook RJ, LaPlante P, Barre P, Cartier P, Fay WP, Goldstein MB, Jindal K, Mandin H, McKenzie JK, Muirhead N, Parfrey PS, Posen GA, Slaughter D, Ulan RA, Werb R. Canadian hemodialysis morbidity study.
Am J Kidney Dis
 
1992
;
XIX
:
214
–34.
19
Pisoni RL, Young EW, Dykstra DM, Greenwood RN, Hecking E, Gillespie B, Wolfe RA, Goodkin DA, Held PJ. Vascular access use in Europe and the United States: Results from the DOPPS.
Kidney Int
 
2002
;
61
:
305
–16.
20
Sehgal AR, Snow RJ, Singer ME, Amini SB, DeOreo PB, Silver MR, Cebul RD. Barriers to adequate delivery of hemodialysis.
Am J Kidney Dis
 
1998
;
31
:
593
–601.
21
Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM. The urea reduction ratio and serum albumin concentration as predictors of mortality in patients undergoing hemodialysis.
N Engl J Med
 
1993
;
329
:
1001
–6.
22
Held PJ, Port FK, Wolfe RA, Stannard DC, Carroll CE, Daugirdas JT, Bloembergen WE, Greer JW, Hakim RM. The dose of hemodialysis and patient mortality.
Kidney Int
 
1996
;
50
:
550
–6.
23
Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE. Hypoalbuminemia, cardiac morbidity, and mortality in end-stage renal disease.
J Am Soc Nephrol
 
1996
;
7
:
728
–36.
24
Holland DC, Lam M. Predictors of hospitalisation and death among pre-dialysis patients: a retrospective cohort study.
Nephrol Dial Transplant
 
2000
;
15
:
650
–8.
25
Becker BN, Coomer RW, Fotiadis C, Evanson J, Shyr Y, Hakim RM. Risk factors for hospitalisation in well-dialysed chronic haemodialysis patients.
Am J Nephrol
 
1999
;
19
:
565
–70.
26
Collins AJ, Li S, Ebben J, Ma JZ, Manning W. Hematocrit levels and associated Medicare expenditures.
Am J Kidney Dis
 
2000
;
36
:
282
–93.

Author notes

From the 1Department of Medicine and Therapeutics, University of Aberdeen, 2Renal Unit, Aberdeen Royal Infirmary, 3Department of Public Health, University of Aberdeen, Aberdeen, and 4Scottish Renal Registry, Glasgow Royal Infirmary, Glasgow, UK